FlowPure: Continuous Normalizing Flows for Adversarial Purification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Collaert, Elias, Rodríguez, Abel, Joos, Sander, Desmet, Lieven, Rimmer, Vera
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918415618277376
author Collaert, Elias
Rodríguez, Abel
Joos, Sander
Desmet, Lieven
Rimmer, Vera
author_facet Collaert, Elias
Rodríguez, Abel
Joos, Sander
Desmet, Lieven
Rimmer, Vera
contents Despite significant advances in the area, adversarial robustness remains a critical challenge in systems employing machine learning models. The removal of adversarial perturbations at inference time, known as adversarial purification, has emerged as a promising defense strategy. To achieve this, state-of-the-art methods leverage diffusion models that inject Gaussian noise during a forward process to dilute adversarial perturbations, followed by a denoising step to restore clean samples before classification. In this work, we propose FlowPure, a novel purification method based on Continuous Normalizing Flows (CNFs) trained with Conditional Flow Matching (CFM) to learn mappings from adversarial examples to their clean counterparts. Unlike prior diffusion-based approaches that rely on fixed noise processes, FlowPure can leverage specific attack knowledge to improve robustness under known threats, while also supporting a more general stochastic variant trained on Gaussian perturbations for settings where such knowledge is unavailable. Experiments on CIFAR-10 and CIFAR-100 demonstrate that our method outperforms state-of-the-art purification defenses in preprocessor-blind and white-box scenarios, and can do so while fully preserving benign accuracy in the former. Moreover, our results show that not only is FlowPure a highly effective purifier but it also holds strong potential for adversarial detection, identifying preprocessor-blind PGD samples with near-perfect accuracy. Our code is publicly available at https://github.com/DistriNet/FlowPure.
format Preprint
id arxiv_https___arxiv_org_abs_2505_13280
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FlowPure: Continuous Normalizing Flows for Adversarial Purification
Collaert, Elias
Rodríguez, Abel
Joos, Sander
Desmet, Lieven
Rimmer, Vera
Machine Learning
Artificial Intelligence
Cryptography and Security
Despite significant advances in the area, adversarial robustness remains a critical challenge in systems employing machine learning models. The removal of adversarial perturbations at inference time, known as adversarial purification, has emerged as a promising defense strategy. To achieve this, state-of-the-art methods leverage diffusion models that inject Gaussian noise during a forward process to dilute adversarial perturbations, followed by a denoising step to restore clean samples before classification. In this work, we propose FlowPure, a novel purification method based on Continuous Normalizing Flows (CNFs) trained with Conditional Flow Matching (CFM) to learn mappings from adversarial examples to their clean counterparts. Unlike prior diffusion-based approaches that rely on fixed noise processes, FlowPure can leverage specific attack knowledge to improve robustness under known threats, while also supporting a more general stochastic variant trained on Gaussian perturbations for settings where such knowledge is unavailable. Experiments on CIFAR-10 and CIFAR-100 demonstrate that our method outperforms state-of-the-art purification defenses in preprocessor-blind and white-box scenarios, and can do so while fully preserving benign accuracy in the former. Moreover, our results show that not only is FlowPure a highly effective purifier but it also holds strong potential for adversarial detection, identifying preprocessor-blind PGD samples with near-perfect accuracy. Our code is publicly available at https://github.com/DistriNet/FlowPure.
title FlowPure: Continuous Normalizing Flows for Adversarial Purification
topic Machine Learning
Artificial Intelligence
Cryptography and Security
url https://arxiv.org/abs/2505.13280